|
主讲人 |
Efang Kong |
简介 |
<p>This paper studies regressions of censored data where both the dependent variable and the censoring variable are assumed to follow multi-index structures as generalization of some parametric or semi-parametric models including the proportional hazard model. Incorporating the idea of redistributionof-mass" (Efron, 1967) for dealing with random censoring, we propose a composite quantile approach that (1) as a general dimension reduction method, can recover the dimension reduction spaces for both the dependent variable and the censoring variable; (2) is computationally straightforward and structure adaptive with better numerical e ciency; (3) runs less risk of model mis-speci cation, yet still retains e ciency comparable to parametric methods such as the Cox proportional hazard model and the accelerated failure time model. Applied in the analysis of the popular primary biliary cirrhosis data, the new approach leads to a revelation more in line with empirical evidence than existing statistical analysis did.</p> |
